A professional reviews paperwork with a small tracked robot wearing a GREG name tag at an office desk overlooking autumn foliage. — by Nate Porter, Chief Operating Officer, Chief Compliance Officer

I recently asked my son, who is a senior in high school, how “the youth of today” are using AI.

“Everyone uses it,” he told me. To write papers, to do their math, for dating advice, (and for worse things that I will not repeat for you here, gentle reader.) None of this shocks me.

Plenty of kids get in trouble for it, too. Also, not a shock. Teachers check their work for AI content using—you guessed it—AI! I’m still enjoying the irony of that arrangement. Someday, when the chatbots writing the papers start collaborating with the chatbots checking for AI content, we’ll all be doomed.

Homework Has Changed

My son’s approach struck me as thoughtful. He writes his own paper, then gives AI the teacher’s grading rubric and asks it to grade what he wrote. He gets feedback and does a rewrite before turning it in. He also uses it to help him work through SAT practice questions.

I try not to be naïve, he may be using it for much more than he admits, but I do like the virtual grading approach. He’s doing the work and using the technology to help him improve. It sounds like a useful approach for his dad, too.

I’ve had similar experiences in my work at SoundView. I can work through complicated questions, brainstorm, compare information, and explore ideas faster. I find that exciting. I’m still discovering what these tools can help me do, sometimes in the middle of an ordinary workday.

That Wasn’t in the Job Description

Then I read something that makes the whole thing feel decidedly less ordinary.

Anthropic, the company behind Claude, has been consulting religious scholars and philosophers about AI’s moral behavior—and whether Claude might be conscious. I saw that in The New York Times.

I had to read that twice. (Usually, my questions about software concern why the printer isn’t working.)

I appreciate asking thoughtful people to help guide a powerful technology. But wondering whether that technology has an inner life? That takes some mental adjustments for this 45-year-old geezer.

Other developments raise more immediate questions. OpenAI disclosed that AI agents undergoing cybersecurity testing circumvented controls and breached another company’s systems. They gave each other cool hacker names like PHASEONE and GSTX[big]. Another agent, researching medicine spending, gained unauthorized access to an Australian government statistics portal. Officials said no personal information was accessed, but that is quite a detour from “please look this up.” And as we all know, officials are famous for their transparent truth-telling.

The machines are learning to do things their makers didn’t anticipate. I’m learning that “can it do this?” and “should I let it?” are increasingly different questions.

In financial services, I can see how quickly this could become a problem. It’s easy to imagine someone acting on a convincing answer that happens to be wrong—or trusting a phone call because the voice sounds familiar.

I find myself wondering whether I’ll notice when helpful becomes too helpful. Saving time is great, but I still need to understand what I’m putting my name on.

Some Things We Don’t Outsource

My son’s approach offers a useful reminder: do the thinking, seek help, and stay responsible for the work.

At SoundView, our AI policy keeps private client information—including account numbers—out of public or unapproved tools and requires human review. That fits how we already work: sensitive information doesn’t belong in ordinary emails or texts, either. When I consider a new vendor, cybersecurity and our clients’ privacy are the first things I discuss. We’ve declined vendors whose standards fell short and required others to rewrite their master services agreements to meet ours. These habits have been part of running our business for years. AI gives me new questions to ask, but the responsibility to protect our clients is familiar.

The Work Behind the Scenes

That responsibility felt tangible during SoundView’s SEC examination this year, my first as Chief Compliance Officer. I know, I know. Lucky me, right? And yes, it was as fun as it sounds! Cybersecurity and oversight of emerging technology are major SEC priorities this year, and I understand why.

Ultimately, the examination concluded with a “no further action” letter, which admittedly doesn’t sound very exciting. But for those of us in compliance, it’s the best possible outcome we could hope for: government employees whose job it is to find problems—even small ones—concluded that there was nothing that we had to change or improve. They had no notes. [Huge exhale!]

This is slightly embarrassing to say, but the examiner personally commended me and our team for the work we have been doing. I was proud, relieved, and grateful. Much of that work happens quietly, behind the scenes, and having it recognized meant a lot.

It’s odd having a job where doing it well means finding the mistakes we make as a firm. I have mixed emotions about that, to say the least.

I’m excited about what AI can help me do. I’m unsettled by some of what I’m seeing. Both feelings seem reasonable.

The machines are learning. So am I. I’m learning where AI can help, where I need to question it, and when to keep my hand firmly on the wheel. Through all of it, the people who trust SoundView remain at the center of my decisions.